Hi-lo guide
Fusion Hi-lo: the edge, step by step, inside a chain
Here, "Fusion" is a page name only; the maths below applies to any hi-lo game built on the same model. The topic is a single piece of arithmetic: how a small edge on each call becomes a larger edge on a whole chain.
One call first
Take one hi-lo call with win chance p. In this model a correct call returns (1 - e) / p times the stake. The expected return is p x (1 - e) / p = 1 - e. With e = 1%, one call returns 0.99 per unit on average, whichever side and whichever face card.
Two calls, then many
In a chain the whole return of the first call rides on the second. The chance of two correct calls is p1 x p2, and the pay is the product of the two step multipliers. Multiply them and every p cancels, leaving (1 - e)2 = 0.9801. Each extra call adds one more factor of (1 - e). After n calls the expected return is (1 - e)n, and the edge on the full chain is 1 - (1 - e)n.
The table below makes this concrete with eight identical calls of higher from a 3. Each step wins 76.92% of the time. Check any row: chain chance x chain multiplier gives 0.99 to the power of the step number.
| Step | Call | Step chance | Chain chance | Chain multiplier |
|---|---|---|---|---|
| 1 | 3 then higher | 76.92% | 76.92% | 1.29x |
| 2 | 3 then higher | 76.92% | 59.17% | 1.66x |
| 3 | 3 then higher | 76.92% | 45.52% | 2.13x |
| 4 | 3 then higher | 76.92% | 35.01% | 2.74x |
| 5 | 3 then higher | 76.92% | 26.93% | 3.53x |
| 6 | 3 then higher | 76.92% | 20.72% | 4.54x |
| 7 | 3 then higher | 76.92% | 15.94% | 5.85x |
| 8 | 3 then higher | 76.92% | 12.26% | 7.53x |
The compounded edge by chain length
| Calls in chain | Expected return per unit | Effective edge on the chain |
|---|---|---|
| 1 | 0.9900 | 1.00% |
| 3 | 0.9703 | 2.97% |
| 5 | 0.9510 | 4.90% |
| 10 | 0.9044 | 9.56% |
| 20 | 0.8179 | 18.21% |
| 50 | 0.6050 | 39.50% |
The effective edge grows a little slower than n x e, because each later step only taxes what survived. Ten calls cost 9.56% rather than a flat 10%. The difference is small for short chains and becomes visible only for long ones.
What the derivation does not say
- It does not depend on which face cards appear. Strong calls and weak calls are taxed at the same rate.
- It does not depend on your choice of side. Each side is priced from its own chance.
- It does not promise any particular result. A chain still either breaks or survives; the formula is an average over many chains.
A related page looks at how the same compounding behaves when the per-call edge is 2% or 3% instead of 1%: see Mystic Hi-lo.